Method and system for realizing processing quality analysis of baby carriage component based on visual inspection

Abstract

The invention relates to the technical field of baby carriage preparation, and discloses a method and system for analyzing the processing quality of baby carriage parts based on visual inspection, and the method comprises the steps: analyzing the quality key attributes corresponding to the baby carriage parts; a part preparation image corresponding to the baby carriage part in the preparation process is collected, a part reference point corresponding to the baby carriage part is set, region extraction processing is carried out on the part preparation image, and a part observation image is obtained; performing optimization processing on the to-be-detected features of the component to obtain target to-be-detected features; calculating the part defect degree corresponding to the baby carriage part; testing the size corresponding to the baby carriage component for multiple times to obtain component size data, determining the actual component size corresponding to the baby carriage component, and calculating the component accuracy corresponding to the baby carriage component; and the component quality corresponding to the baby carriage component is evaluated, and a processing quality result corresponding to the baby carriage component is generated. The invention mainly aims to solve the problem that the quality analysis result of the baby carriage component is inaccurate.

Description

Technical Field

[0001] The invention relates to a method and system for analyzing the processing quality of baby carriage parts based on visual inspection, and belongs to the technical field of baby carriage preparation. Background Art

[0002] Automated baby stroller parts processing quality analysis is crucial in baby stroller production. The quality of baby stroller parts is directly related to the safety and service life of the baby stroller. With the increasingly fierce competition in the baby stroller market, consumers have higher and higher requirements for the quality of baby strollers.

[0003] The existing quality analysis of baby stroller parts processing mainly relies on manual sampling. Workers randomly select baby stroller parts from the production line and inspect the appearance and size of the parts by visual observation and simple measuring tools. For example, check whether the metal pipes of the baby stroller have obvious scratches and bends, and measure whether the length and diameter of the pipes are within the specified range; for the plastic shell of the baby stroller, check whether its color is uniform, whether it is deformed, and whether the injection port is flat. This manual sampling method has many limitations. First, the speed of manual inspection is slow, which is difficult to meet the rapid inspection needs of large-scale production, resulting in limited production efficiency. Secondly, manual inspection is easily affected by subjective factors, and different workers may have different grasps of quality standards, resulting in inaccurate quality analysis results of baby stroller parts. Moreover, manual inspection can only inspect the sampled parts, and it is impossible to achieve full inspection of each production part. Parts with quality problems may be missed, thereby affecting the quality of the entire batch of baby strollers. Therefore, a method that can improve the analysis accuracy of the processing quality of baby stroller parts is needed. Summary of the invention

[0004] The present invention provides a method and system for analyzing the processing quality of baby carriage parts based on visual inspection, the main purpose of which is to solve the problem of inaccurate inspection results.

[0005] To achieve the above-mentioned purpose, the present invention provides a method for analyzing the processing quality of baby carriage parts based on visual inspection, comprising: Obtaining a manufacturing process step corresponding to a baby stroller component, querying a manufacturing criterion corresponding to the manufacturing process step, and analyzing a key quality attribute corresponding to the baby stroller component based on the manufacturing criterion; Collecting component preparation images corresponding to the baby stroller components in the preparation process, setting component reference points corresponding to the baby stroller components based on the key quality attributes, and performing region extraction processing on the component preparation images in combination with the component reference points to obtain component observation images; Extracting the component features to be measured corresponding to the baby carriage component from the component observation image, and performing optimal processing on the component features to be measured to obtain target features to be measured; Retrieving a component reference image corresponding to the baby stroller component, and extracting a reference feature to be measured from the component reference image based on the component reference point, and calculating the component defect degree corresponding to the baby stroller component by combining the target feature to be measured and the reference feature to be measured, wherein calculating the component defect degree corresponding to the baby stroller component by combining the target feature to be measured and the reference feature to be measured includes: Performing vectorization processing on the target feature to be measured and the reference feature to be measured respectively to obtain a first feature vector and a second feature vector; Calculate the feature deviation between the target feature to be measured and the reference feature to be measured by combining the first feature vector and the second feature vector; The feature weight corresponding to the target feature to be measured is calculated, and the component defect degree corresponding to the baby carriage component is calculated by combining the feature weight and the feature deviation through the following formula: Among them, E represents the component defect degree corresponding to the baby stroller component, It represents the feature weight corresponding to the ath feature in the target feature to be tested. It represents the feature deviation corresponding to the ath feature in the target feature to be tested, a represents the feature sequence number of the target feature to be tested, r represents the number of features of the target feature to be tested, Indicates the maximum value of feature deviation; Perform multiple tests on the dimensions corresponding to the baby stroller components to obtain component dimension data, determine the actual dimensions of the components corresponding to the baby stroller components based on the component dimension data, and calculate the component accuracy corresponding to the baby stroller components by combining the actual dimensions of the components and the component dimension data; The component quality corresponding to the baby carriage component is evaluated in combination with the component defect degree and the component accuracy, and the processing quality result corresponding to the baby carriage component is generated based on the component quality.

[0006] Optionally, analyzing the key quality attributes corresponding to the baby carriage components based on the manufacturing criteria includes: Performing text recognition on the preparation criteria to obtain a criteria text; Performing semantic analysis on the standard text to obtain the standard text semantics; Extracting key criterion text from the criterion text based on the semantics of the criterion text; Analyzing the criterion relevance between the key criterion texts, and extracting criterion labels corresponding to the key criterion texts; Performing attribute parsing on the key criterion text to obtain criterion attributes; The key quality attributes corresponding to the baby carriage components are extracted from the criterion attributes in combination with the criterion labels and the criterion associations.

[0007] Optionally, analyzing the criterion relevance between the key criterion texts includes: Extracting key terms from the key criteria text, and standardizing the key terms to obtain standard key terms; Defining correlation factors in the standard key terms, and constructing a criterion correlation matrix corresponding to the key criterion text based on the correlation factors; Based on the criterion association matrix, calculating the criterion association degree between the key criterion texts; According to the criterion relevance, the criterion relevance between the key criterion texts is analyzed.

[0008] Optionally, combining the component reference point and performing region extraction processing on the component preparation image to obtain the component observation image includes: In combination with the component reference point, performing region marking processing on the component preparation image to obtain a component marking preparation image; Performing region segmentation processing on the component marking preparation image to obtain a component region image; Performing noise reduction processing on the component region image to obtain a noise-reduced component image; The noise-reduced component image is digitally quantized to obtain a component observation image.

[0009] Optionally, the performing digital quantization processing on the noise-reduced component image to obtain the component observation image includes: Constructing an image coordinate system corresponding to the noise reduction component image, and identifying a pixel coordinate value of each pixel in the noise reduction component image in the image coordinate system; Based on the pixel coordinate values, respectively calculating the pixel grayscale change rate and the change rate direction corresponding to each pixel in the noise reduction component image; Based on the pixel grayscale change rate and the change rate direction, edge extraction is performed on the denoised component image to obtain the component image edge; Perform edge fitting processing on the edge of the component image to obtain a component observation image.

[0010] Optionally, the calculating, based on the pixel coordinate values, the pixel grayscale change rate and the change rate direction corresponding to each pixel in the noise reduction component image respectively includes: Based on the pixel coordinate values, calculating the horizontal grayscale change rate and the vertical grayscale change rate corresponding to each pixel in the noise reduction component image; Combining the horizontal grayscale change rate and the vertical grayscale change rate, the pixel grayscale change rate corresponding to each pixel in the noise reduction component image can be calculated by the following formula: Among them, B represents the grayscale change rate of each pixel in the noise reduction component image. Indicates the horizontal grayscale change rate, Indicates the vertical grayscale change rate; Combining the horizontal grayscale change rate and the vertical grayscale change rate, the change rate direction corresponding to each pixel in the noise reduction component image can be calculated by the following formula: Where D represents the direction of the rate of change corresponding to each pixel in the denoising component image, represents the inverse tangent function, Indicates the horizontal grayscale change rate, Indicates the vertical grayscale change rate.

[0011] Optionally, extracting the component features to be measured corresponding to the baby carriage components from the component observation image includes: Performing contrast enhancement processing on the component observation image to enhance the observation image, and extracting observation color features corresponding to the enhanced observation image; Performing grayscale processing on the enhanced observation image to obtain a grayscale observation image, and extracting observation texture features corresponding to the grayscale observation image; Performing boundary detection on the enhanced observation image to obtain an observation boundary, and extracting boundary shape features corresponding to the observation boundary; The observed color features, the observed texture features and the boundary shape features are combined to generate component features to be measured corresponding to the baby carriage components.

[0012] Optionally, the performing optimal processing on the feature to be measured of the component to obtain a target feature to be measured includes: Performing threshold segmentation processing on the features to be tested of the component to obtain segmented features to be tested; Performing feature ablation processing on the segmented features to be measured to obtain ablated features to be measured; The characteristic Gini index corresponding to the ablation feature to be measured is calculated, and based on the characteristic Gini index, the ablation feature to be measured is optimized to obtain a target feature to be measured.

[0013] Optionally, the calculating the component accuracy corresponding to the baby carriage component by combining the actual size of the component and the component size data includes: The size mean corresponding to the component size data is calculated, and the size standard deviation of the baby stroller component in multiple tests is calculated by combining the size mean and the component size data using the following formula: Where H represents the standard deviation of the dimensions of the stroller parts in multiple tests. Indicates the size value corresponding to the i-th data in the component size data, i indicates the serial number corresponding to the component size data, and n indicates the number of data corresponding to the component size data. represents the mean value of size; Calculating the size tolerance corresponding to the stroller component based on the actual size of the component and the preset basic size of the component; Combining the size tolerance, the size mean, the actual size of the component and the size standard deviation, the component accuracy corresponding to the stroller component is calculated by the following formula: Among them, K represents the component accuracy corresponding to the baby stroller parts, represents the size tolerance of the bth component in the baby carriage. represents the mean size, Indicates the actual size of the bth component in the baby carriage. represents the size standard deviation of the bth component in the baby stroller, and b represents the serial number of the baby stroller component.

[0014] A system for analyzing the processing quality of baby carriage parts based on visual inspection is characterized in that the system comprises: A quality key attribute analysis module, used to obtain the manufacturing process steps corresponding to the baby stroller components, query the manufacturing criteria corresponding to the manufacturing process steps, and analyze the quality key attributes corresponding to the baby stroller components based on the manufacturing criteria; An image processing module, for collecting component preparation images corresponding to the baby stroller components in the preparation process, setting component reference points corresponding to the baby stroller components based on the key quality attributes, and performing region extraction processing on the component preparation images in combination with the component reference points to obtain component observation images; A feature optimization processing module is used to extract the component features to be measured corresponding to the baby carriage components from the component observation image, and perform optimization processing on the component features to be measured to obtain target features to be measured; A component defect calculation module is used to retrieve a component reference image corresponding to the baby stroller component, and based on the component reference point, extract a reference feature to be measured from the component reference image, and calculate the component defect corresponding to the baby stroller component by combining the target feature to be measured and the reference feature to be measured, wherein the calculation of the component defect corresponding to the baby stroller component by combining the target feature to be measured and the reference feature to be measured includes: Performing vectorization processing on the target feature to be measured and the reference feature to be measured respectively to obtain a first feature vector and a second feature vector; Calculate the feature deviation between the target feature to be measured and the reference feature to be measured by combining the first feature vector and the second feature vector; The feature weight corresponding to the target feature to be measured is calculated, and the component defect degree corresponding to the baby carriage component is calculated by combining the feature weight and the feature deviation through the following formula: Among them, E represents the component defect degree corresponding to the baby stroller component, It represents the feature weight corresponding to the ath feature in the target feature to be tested. It represents the feature deviation corresponding to the ath feature in the target feature to be tested, a represents the feature sequence number of the target feature to be tested, r represents the number of features of the target feature to be tested, Indicates the maximum value of feature deviation; A component accuracy calculation module, used for performing multiple tests on the dimensions corresponding to the components of the baby carriage to obtain component dimension data, determining the actual dimensions of the components corresponding to the components of the baby carriage based on the component dimension data, and calculating the component accuracy corresponding to the components of the baby carriage by combining the actual dimensions of the components and the component dimension data; The component quality evaluation module is used to evaluate the component quality corresponding to the baby carriage component in combination with the component defect degree and the component accuracy, and generate a processing quality result corresponding to the baby carriage component based on the component quality.

[0015] Compared with the problems described in the background technology, the present invention analyzes the key quality attributes corresponding to the baby stroller components based on the preparation criteria, and can obtain the unique properties of the baby stroller components related to production quality, understand the quality level corresponding to the baby stroller components, and thus lay a foundation for the subsequent setting of component reference points corresponding to the baby stroller components. The present invention sets the component reference points corresponding to the baby stroller components based on the key quality attributes, so as to facilitate the subsequent precise positioning of key areas from the component preparation images to improve the accuracy of the quality analysis of the baby stroller components. The present invention extracts the component features to be tested corresponding to the baby stroller components from the component observation images, and can obtain specific attributes that need to be tested in the baby stroller components, thereby providing an accurate basis for the quality inspection of the baby stroller components. The present invention calculates the component defects corresponding to the baby stroller components by combining the target features to be tested and the benchmark features to be tested. The degree of defects of the baby stroller parts in the production process can be understood through the defect degree of the parts, thereby improving the accuracy of subsequent component quality evaluation of the baby stroller parts. The present invention determines the actual size of the parts corresponding to the baby stroller parts based on the component size data, and can obtain detailed size information corresponding to the baby stroller parts. The accuracy of the parts corresponding to the baby stroller parts can be calculated by combining the actual size of the parts and the component size data. The processing accuracy of the baby stroller parts in production can be understood through the component accuracy, which lays a foundation for the evaluation of the component quality corresponding to the baby stroller parts. The present invention can improve the accuracy of the component quality evaluation of the baby stroller parts by combining the component defect degree and the component accuracy, thereby obtaining the accurate quality results corresponding to the baby stroller parts in production and processing, thereby achieving the purpose of improving the accuracy of the processing quality analysis of the baby stroller parts. Therefore, the present invention proposes a method and system for analyzing the processing quality of baby stroller parts based on visual inspection, so as to improve the analysis accuracy of the processing quality of baby stroller parts. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A schematic diagram of a process flow of a method for analyzing the processing quality of baby carriage components based on visual inspection provided by an embodiment of the present invention; Figure 2 A functional module diagram of a system for analyzing the processing quality of baby carriage components based on visual inspection provided in one embodiment of the present invention.

[0017] The purpose, features and advantages of the present invention will be further described with reference to the accompanying drawings in conjunction with the embodiments. DETAILED DESCRIPTION

[0018] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.

[0019] The embodiment of the present application provides a method for analyzing the processing quality of baby stroller parts based on visual inspection. The execution subject of the method for analyzing the processing quality of baby stroller parts based on visual inspection includes but is not limited to at least one of the electronic devices such as a server and a terminal that can be configured to execute the method provided by the embodiment of the present application. In other words, the method for analyzing the processing quality of baby stroller parts based on visual inspection can be executed by software or hardware installed in a terminal device or a server device. The server includes but is not limited to: a single server, a server cluster, a cloud server or a cloud server cluster, etc.

[0020] Embodiment 1: Reference Figure 1 FIG. 1 is a flow chart of a method for analyzing the processing quality of baby stroller parts based on visual inspection according to an embodiment of the present invention. In this embodiment, the method for analyzing the processing quality of baby stroller parts based on visual inspection includes: S1. Obtain a manufacturing process step corresponding to a baby stroller component, query a manufacturing criterion corresponding to the manufacturing process step, and analyze a key quality attribute corresponding to the baby stroller component based on the manufacturing criterion.

[0021] The present invention is based on the preparation criteria and analyzes the key quality attributes corresponding to the baby stroller components, so as to obtain the unique properties of the baby stroller components related to production quality, understand the quality levels corresponding to the baby stroller components, and thus lay a foundation for the subsequent setting of component reference points corresponding to the baby stroller components, wherein the baby stroller components are the various components that constitute the baby stroller, the preparation process steps are the production operation steps corresponding to the baby stroller components, the preparation criteria are the standards that regulate and guide the preparation process steps, and the key quality attributes are the core characteristics that determine the quality of the baby stroller components. Furthermore, the acquisition of the preparation process steps corresponding to the baby stroller components can be achieved through human-computer interaction; the query of the preparation criteria corresponding to the preparation process steps can be achieved through the production management system corresponding to the baby stroller components.

[0022] In detail, the analyzing the key quality attributes corresponding to the baby carriage components based on the manufacturing criteria includes: Performing text recognition on the preparation criteria to obtain a criteria text; Performing semantic analysis on the standard text to obtain the standard text semantics; Extracting key criterion text from the criterion text based on the semantics of the criterion text; Analyzing the criterion relevance between the key criterion texts, and extracting criterion labels corresponding to the key criterion texts; Performing attribute parsing on the key criterion text to obtain criterion attributes; The key quality attributes corresponding to the baby carriage components are extracted from the criterion attributes in combination with the criterion labels and the criterion associations.

[0023] It should be explained that the criterion text is the text information in the preparation criterion, the criterion text semantics is the interpretation of the meaning corresponding to the criterion text, the key criterion text is the representative text in the criterion text, the criterion relevance represents the association relationship between the key criterion texts, the criterion label is the classification identifier corresponding to the key criterion text, and the criterion attribute is the property corresponding to the key criterion text.

[0024] Furthermore, the text recognition of the preparation criteria can be achieved through OCR recognition technology; the semantic parsing of the criteria text can be achieved through semantic parsing method; the extraction of key criteria text in the criteria text can be achieved through TF-IDF algorithm; the extraction of criteria labels corresponding to the key criteria text can be achieved through natural language processing tools, such as NLTK (Natural Language Toolkit) tool; the attribute parsing of the key criteria text can be achieved through grammatical rule analysis method, for example, for the sentence "the diameter of the baby stroller wheel should be between 20-25 cm", the attributes of the text are parsed by identifying the subject (baby stroller wheel), predicate (should be between...) and object (diameter and its range 20-25 cm) in the sentence, where the attributes include objects (baby stroller wheels), attribute features (diameter) and attribute values ​​(20-25 cm), quality labels are identified from the criteria labels, and labels associated with the quality labels are determined in combination with the criteria relevance, so as to extract the key quality attributes corresponding to the baby stroller components from the criteria attributes.

[0025] Further, as an optional embodiment of the present invention, the analyzing the criterion relevance between the key criterion texts includes: Extracting key terms from the key criteria text, and standardizing the key terms to obtain standard key terms; Defining correlation factors in the standard key terms, and constructing a criterion correlation matrix corresponding to the key criterion text based on the correlation factors; Based on the criterion association matrix, calculating the criterion association degree between the key criterion texts; According to the criterion relevance, the criterion relevance between the key criterion texts is analyzed.

[0026] It should be explained that the key terms are professional descriptions in the key criteria texts. For example, the terms "frame material" and "tensile strength" are extracted from "the tensile strength of the frame material is not less than 10MPa". These terms are the basis for understanding the content of the criteria and analyzing the correlation. The standard key terms are the terms obtained after the key terms are unified. If in different criteria texts, some describe the force bearing capacity of the frame as "tensile strength" and some describe it as "tensile strength", they need to be unified into a standard term, such as "tensile strength". The correlation factors are factors with common quality attributes in the standard key terms, such as whether the same baby carriage parts are involved, whether the same quality attributes are affected, whether they are in the same production link, etc. The criterion correlation matrix is ​​a square matrix composed of the correlation factors, and the criterion correlation degree indicates the degree of correlation between the key criterion texts.

[0027] Furthermore, the extraction of key terms in the key criterion text can be implemented through relevant libraries in programming languages, such as Scikit-learn; the definition of associated factors in the standard key terms can be implemented through the expert knowledge method, relying on the knowledge and experience of domain experts to define associated factors. In the baby stroller industry, experts are very familiar with the various components, production processes, quality standards, etc. of baby strollers. For example, for the standard key term "baby stroller frame", experts can determine the factors related to it based on their professional knowledge, such as "material (metal or plastic)", "structural strength", "welding process (if any)", etc.; the construction of the criterion association matrix corresponding to the key criterion text can be implemented through a matrix function, such as the zero matrix function; the matrix similarity between the criterion association matrices is calculated, and the criterion association between the key criterion texts is obtained based on the matrix similarity; based on the criterion association, the criterion association between the key criterion texts is analyzed. If the criterion association is greater than the preset association, it means that the criterion association is high, and the criterion association between the key criterion texts is high. The preset association can be set according to the actual application scenario.

[0028] S2. Collect component preparation images corresponding to the baby stroller components in the preparation process step, set component reference points corresponding to the baby stroller components based on the key quality attributes, and perform region extraction processing on the component preparation images in combination with the component reference points to obtain component observation images.

[0029] The present invention sets the component reference points corresponding to the baby stroller components based on the key quality attributes, so as to facilitate the subsequent precise positioning of key areas from the component preparation images, so as to improve the accuracy of the quality analysis of the baby stroller components. It should be explained that the component preparation image is the image corresponding to the baby stroller components in the preparation process, and the component reference points are the key position identifiers of the baby stroller components. Furthermore, the collection of the component preparation images corresponding to the baby stroller components in the preparation process can be achieved by an industrial camera; the position of the key quality attributes in the baby stroller components is determined, and the position is used as the component reference point corresponding to the baby stroller components.

[0030] The present invention combines the component reference point to perform region extraction processing on the component preparation image, thereby removing images of irrelevant regions in the component preparation image, improving the image quality and reducing the difficulty of subsequent image processing. It should be explained that the component observation image is the image obtained after removing images of irrelevant regions in the component preparation image.

[0031] Specifically, the step of combining the component reference point and performing region extraction processing on the component preparation image to obtain the component observation image includes: In combination with the component reference point, performing region marking processing on the component preparation image to obtain a component marking preparation image; Performing region segmentation processing on the component marking preparation image to obtain a component region image; Performing noise reduction processing on the component region image to obtain a noise-reduced component image; The noise-reduced component image is digitally quantized to obtain a component observation image.

[0032] It should be explained that the component marking preparation image is an image obtained after the component reference point is marked on the component preparation image, the component area image is an image obtained after the image outside the component reference point is removed from the component marking preparation image, and the denoised component image is an image obtained after the noise interference in the component area image is removed.

[0033] Furthermore, the area marking processing of the component preparation image can be achieved by manual marking; the area segmentation processing of the component marking preparation image can be achieved by a segmentation tool, and the segmentation tool is compiled by JAVA language; the noise reduction processing of the component area image can be achieved by a low-pass filter.

[0034] Further, as an optional embodiment of the present invention, the digital quantization processing of the noise-reduced component image to obtain the component observation image includes: Constructing an image coordinate system corresponding to the noise reduction component image, and identifying a pixel coordinate value of each pixel in the noise reduction component image in the image coordinate system; Based on the pixel coordinate values, respectively calculating the pixel grayscale change rate and the change rate direction corresponding to each pixel in the noise reduction component image; Based on the pixel grayscale change rate and the change rate direction, edge extraction is performed on the denoised component image to obtain the component image edge; Perform edge fitting processing on the edge of the component image to obtain a component observation image.

[0035] It should be explained that the pixel coordinate value is a description of the specific position corresponding to each pixel in the denoising component image in the image coordinate system, the pixel grayscale change rate and the change rate direction are respectively the change speed and the change direction corresponding to each pixel in the denoising component image, and the component image edge is the edge position in the denoising component image. Further, the construction of the image coordinate system corresponding to the denoising component image can be achieved through the drawing function in the OpenCV library, for example, by specifying the origin position (usually in the upper left corner of the image), and then defining the directions of the x-axis and the y-axis (the x-axis is usually horizontal to the right, and the y-axis is usually vertically downward), using the line function to draw the coordinate axis, and then using the putText function to mark the coordinate axis scale; combining the numerical values ​​corresponding to the pixel grayscale change rate and the change rate direction, the edge of the denoising component image is extracted to obtain the component image edge. If the pixel grayscale change rate and the numerical values ​​of the change rate direction change rapidly, it means that the corresponding pixel is an edge pixel, and the component image edge is constructed according to the edge pixels; the edge fitting processing of the component image edge can be achieved by the least squares fitting method.

[0036] Further, as an optional embodiment of the present invention, the step of calculating the pixel grayscale change rate and the change rate direction corresponding to each pixel in the noise reduction component image based on the pixel coordinate value includes: Based on the pixel coordinate values, calculating the horizontal grayscale change rate and the vertical grayscale change rate corresponding to each pixel in the noise reduction component image; Combining the horizontal grayscale change rate and the vertical grayscale change rate, the pixel grayscale change rate corresponding to each pixel in the noise reduction component image can be calculated by the following formula: Among them, B represents the grayscale change rate of each pixel in the noise reduction component image. Indicates the horizontal grayscale change rate, Indicates the vertical grayscale change rate; Combining the horizontal grayscale change rate and the vertical grayscale change rate, the change rate direction corresponding to each pixel in the noise reduction component image can be calculated by the following formula: Where D represents the direction of the rate of change corresponding to each pixel in the denoising component image, represents the inverse tangent function, Indicates the horizontal grayscale change rate, Indicates the vertical grayscale change rate.

[0037] It should be explained that the horizontal grayscale change rate and the vertical grayscale change rate respectively represent the grayscale change degree corresponding to each pixel in the denoising component image in the horizontal direction and the vertical direction, the pixel grayscale change rate represents the comprehensive grayscale change degree corresponding to each pixel in the denoising component image, and the change rate direction represents the direction of the grayscale change corresponding to each pixel in the denoising component image. Furthermore, the calculation of the horizontal grayscale change rate and the vertical grayscale change rate corresponding to each pixel in the denoising component image can be achieved by the Sobel operator.

[0038] S3. Extracting component features to be measured corresponding to the baby carriage components from the component observation image, and performing optimal processing on the component features to be measured to obtain target features to be measured.

[0039] The present invention can obtain the specific attributes that need to be tested in the baby stroller components by extracting the component features to be tested corresponding to the baby stroller components from the component observation image, thereby providing an accurate basis for the quality inspection of the baby stroller components. It should be explained that the component features to be tested are the specific attributes corresponding to the baby stroller components in the component observation image, such as color features and texture features.

[0040] In detail, the extracting the component features to be measured corresponding to the baby carriage components from the component observation image includes: Performing contrast enhancement processing on the component observation image to enhance the observation image, and extracting observation color features corresponding to the enhanced observation image; Performing grayscale processing on the enhanced observation image to obtain a grayscale observation image, and extracting observation texture features corresponding to the grayscale observation image; Performing boundary detection on the enhanced observation image to obtain an observation boundary, and extracting boundary shape features corresponding to the observation boundary; The observed color features, the observed texture features and the boundary shape features are combined to generate component features to be measured corresponding to the baby carriage components.

[0041] It should be explained that the enhanced observation image is the image obtained after the component observation image is optimized, the observed color feature is the color attribute representation corresponding to the enhanced observation image, the grayscale observation image is the image represented by only a single color, the observed texture feature is the texture attribute representation corresponding to the grayscale observation image, the observed boundary is the shape boundary in the enhanced observation image, and the boundary shape feature is the shape-specific attribute corresponding to the observed boundary.

[0042] Furthermore, the contrast enhancement processing of the component observation image can be achieved through a histogram equalization method; the extraction of the observed color features corresponding to the enhanced observation image can be achieved through a color histogram method; the grayscale processing of the enhanced observation image can be achieved through an average method, averaging the values ​​of the three channels R, G, and B of the color image to obtain the grayscale value; the extraction of the observed texture features corresponding to the grayscale observation image can be achieved through a grayscale co-occurrence matrix algorithm; the boundary detection of the enhanced observation image can be achieved through a boundary detection algorithm, such as the Prewitt operator; the extraction of the boundary shape features corresponding to the observed boundary can be achieved through a shape feature extraction algorithm, such as SIFT (Scale Invariant Feature Transform), SURF (Speeded Up Robust Features) and other algorithms.

[0043] The present invention can remove redundant features in the features to be measured of the component by performing optimal processing on the features to be measured of the component, thereby retaining the most representative features in the features to be measured of the component, and improving the calculation accuracy of the component defect degree corresponding to the subsequent baby carriage component. It should be explained that the target feature to be measured is the feature obtained after removing the redundant features in the features to be measured of the component.

[0044] Specifically, the optimal processing of the component features to be measured to obtain the target features to be measured includes: Performing threshold segmentation processing on the features to be tested of the component to obtain segmented features to be tested; Performing feature ablation processing on the segmented features to be measured to obtain ablated features to be measured; The characteristic Gini index corresponding to the ablation feature to be measured is calculated, and based on the characteristic Gini index, the ablation feature to be measured is optimized to obtain a target feature to be measured.

[0045] It should be explained that the segmented features to be tested are the features to be tested of the component divided into different categories, the ablation features to be tested are the features with specific meanings obtained after removing some features from the segmented features to be tested, and the feature Gini index represents the feature purity corresponding to the ablation features to be tested. The smaller the feature Gini index, the higher the corresponding feature purity, and the more important the corresponding ablation features to be tested.

[0046] Furthermore, the threshold segmentation processing of the component features to be tested can be achieved by a fixed threshold-based method, and the threshold is manually determined according to experience, domain knowledge or specific standards. For example, when detecting the degree of wear of a mechanical component, if it is known that when the thickness of the component is worn to less than 5mm, its performance will be significantly affected, then 5mm is a manually set threshold. For the measured component thickness data, those less than 5mm are classified as high wear categories, and those greater than or equal to 5mm are classified as low wear categories; the feature ablation processing of the segmented features to be tested can be achieved by a feature corrosion operation; the calculation of the feature Gini index corresponding to the ablated feature to be tested can be achieved by a Gini index function, and the Gini index function is compiled by a scripting language, such as JS scripting language; when the feature Gini index is greater than the average Gini index, the ablated feature to be tested is optimized to obtain the target feature to be tested.

[0047] S4, retrieving a component reference image corresponding to the baby stroller component, and extracting a reference feature to be measured from the component reference image based on the component reference point, and calculating a component defect degree corresponding to the baby stroller component by combining the target feature to be measured and the reference feature to be measured.

[0048] The present invention calculates the component defect degree corresponding to the baby stroller component by combining the target feature to be measured and the reference feature to be measured. The component defect degree can be used to understand the defect degree of the baby stroller component in the production process, thereby improving the accuracy of subsequent component quality evaluation corresponding to the baby stroller component. It should be explained that the component reference image is the standard reference image corresponding to the baby stroller component, and the reference feature to be measured is the standard reference feature of the component reference image corresponding to the baby stroller component. The component defect degree represents the defect degree corresponding to the baby stroller component. Furthermore, the component reference image corresponding to the baby stroller component can be obtained from a cloud database, which is a storage database established by a quality control department, and the standard design drawing of each baby stroller component is stored in the database as a reference image. The extraction principle of the reference feature to be measured is the same as that of the target feature to be measured, which will not be elaborated here.

[0049] In detail, the step of combining the target feature to be measured and the reference feature to be measured to calculate the component defect degree corresponding to the baby carriage component includes: Performing vectorization processing on the target feature to be measured and the reference feature to be measured respectively to obtain a first feature vector and a second feature vector; Calculate the feature deviation between the target feature to be measured and the reference feature to be measured by combining the first feature vector and the second feature vector; The feature weight corresponding to the target feature to be measured is calculated, and the component defect degree corresponding to the baby carriage component is calculated by combining the feature weight and the feature deviation through the following formula: Among them, E represents the component defect degree corresponding to the baby stroller component, It represents the feature weight corresponding to the ath feature in the target feature to be tested. It represents the feature deviation corresponding to the ath feature in the target feature to be tested, a represents the feature sequence number of the target feature to be tested, r represents the number of features of the target feature to be tested, Indicates the maximum value of feature deviation.

[0050] It should be explained that the first feature vector and the second feature vector are respectively the expression vectors corresponding to the target feature to be tested and the benchmark feature to be tested, the feature deviation represents the degree of difference between the target feature to be tested and the benchmark feature to be tested, and the feature weight represents the feature importance corresponding to the target feature to be tested. Furthermore, the vectorization processing of the target feature to be tested and the benchmark feature to be tested can be achieved through a bag-of-words model; the feature deviation between the target feature to be tested and the benchmark feature to be tested can be obtained by calculating the difference between the first feature vector and the second feature vector; the calculation of the feature weight corresponding to the target feature to be tested can be achieved through an information gain algorithm.

[0051] S5. Perform multiple tests on the dimensions corresponding to the baby stroller components to obtain component dimension data, determine the actual dimensions of the components corresponding to the baby stroller components based on the component dimension data, and calculate the component accuracy corresponding to the baby stroller components by combining the actual dimensions of the components and the component dimension data.

[0052] The present invention can obtain detailed size information corresponding to the baby stroller component by determining the actual size of the component corresponding to the baby stroller component based on the component size data, and calculate the component accuracy corresponding to the baby stroller component by combining the actual size of the component and the component size data. The component accuracy can be used to understand the processing accuracy of the baby stroller component in production, which lays a foundation for the evaluation of the component quality corresponding to the baby stroller component. It should be explained that the component size data is the size numerical information corresponding to the baby stroller component, the actual size of the component is the real size value corresponding to the baby stroller component, and the component accuracy is the processing accuracy corresponding to the baby stroller component. Furthermore, multiple tests of the size corresponding to the baby stroller component can be achieved by measuring tools, such as a vernier caliper tool; the actual size of the component corresponding to the baby stroller component can be obtained by the value recorded in the component size data.

[0053] Specifically, the calculating the component accuracy corresponding to the baby carriage component by combining the actual size of the component and the component size data includes: The size mean corresponding to the component size data is calculated, and the size standard deviation of the baby stroller component in multiple tests is calculated by combining the size mean and the component size data using the following formula: Where H represents the standard deviation of the dimensions of the stroller parts in multiple tests. Indicates the size value corresponding to the i-th data in the component size data, i indicates the serial number corresponding to the component size data, and n indicates the number of data corresponding to the component size data. represents the mean value of size; Calculating the size tolerance corresponding to the stroller component based on the actual size of the component and the preset basic size of the component; Combining the size tolerance, the size mean, the actual size of the component and the size standard deviation, the component accuracy corresponding to the stroller component is calculated by the following formula: Among them, K represents the component accuracy corresponding to the baby stroller parts, represents the size tolerance of the bth component in the baby carriage. represents the mean size, Indicates the actual size of the bth component in the baby carriage. represents the size standard deviation of the bth component in the baby stroller, and b represents the serial number of the baby stroller component.

[0054] It should be explained that the dimensional standard deviation indicates the degree of dimensional dispersion of the baby stroller components in multiple tests, the preset basic dimensions of the components are the standard dimensional values ​​corresponding to the baby stroller components, and the dimensional tolerance indicates the allowable dimensional error corresponding to the baby stroller components. Furthermore, the dimensional tolerance corresponding to the baby stroller components can be obtained by calculating the difference between the actual dimensions of the components and the preset basic dimensions of the components.

[0055] S6. Evaluate the component quality corresponding to the baby carriage component in combination with the component defect degree and the component accuracy, and generate a processing quality result corresponding to the baby carriage component based on the component quality.

[0056] The present invention can improve the evaluation accuracy of the component quality corresponding to the baby stroller component by combining the component defect degree and the component accuracy, and then obtain the accurate quality results corresponding to the baby stroller component in the production and processing, so as to achieve the purpose of improving the accuracy of the processing quality analysis of the baby stroller component. It should be explained that the component quality is the component performance corresponding to the baby stroller component. Further, in combination with the component defect degree and the component accuracy, the component quality corresponding to the baby stroller component is evaluated. For example, for the wheels of a baby stroller, if the component accuracy is high, it means that the size and shape of the wheel accurately meet the design requirements, such as the diameter deviation is extremely small. However, if the component defect degree is high, it may be that there are cracks on the wheel surface or uneven material. On the whole, although the size is accurate, there are defects. The component quality corresponding to the baby stroller component is unqualified. For example, for a baby stroller frame, if the component accuracy is low and the size does not meet the standard, even if there are no obvious defects on the surface, the quality of the frame does not meet the requirements, because inaccurate dimensions will affect the assembly and overall structural stability of the baby stroller. When the component defectivity and the component accuracy are both high, the component quality corresponding to the baby stroller component is qualified. The component defectivity and the component accuracy can be compared with the corresponding preset thresholds. For example, the preset threshold can be set to 0.8. When the component defectivity and the component accuracy are both greater than 0.8, it means that the component defectivity and the component accuracy are both high. Based on the component quality, the processing quality results corresponding to the baby stroller components can be generated by quality tools, such as the cause-effect diagram method.

[0057] Compared with the problems described in the background technology, the present invention analyzes the key quality attributes corresponding to the baby stroller components based on the preparation criteria, and can obtain the unique properties of the baby stroller components related to production quality, understand the quality level corresponding to the baby stroller components, and thus lay a foundation for the subsequent setting of component reference points corresponding to the baby stroller components. The present invention sets the component reference points corresponding to the baby stroller components based on the key quality attributes, so as to facilitate the subsequent precise positioning of key areas from the component preparation images to improve the accuracy of the quality analysis of the baby stroller components. The present invention extracts the component features to be tested corresponding to the baby stroller components from the component observation images, and can obtain specific attributes that need to be tested in the baby stroller components, thereby providing an accurate basis for the quality inspection of the baby stroller components. The present invention calculates the component defects corresponding to the baby stroller components by combining the target features to be tested and the benchmark features to be tested. The degree of defects of the baby stroller parts in the production process can be understood through the defect degree of the parts, thereby improving the accuracy of subsequent component quality evaluation of the baby stroller parts. The present invention determines the actual size of the parts corresponding to the baby stroller parts based on the component size data, and can obtain detailed size information corresponding to the baby stroller parts. The accuracy of the parts corresponding to the baby stroller parts can be calculated by combining the actual size of the parts and the component size data. The processing accuracy of the baby stroller parts in production can be understood through the component accuracy, which lays a foundation for the evaluation of the component quality corresponding to the baby stroller parts. The present invention can improve the accuracy of the component quality evaluation of the baby stroller parts by combining the component defect degree and the component accuracy, thereby obtaining the accurate quality results corresponding to the baby stroller parts in production and processing, thereby achieving the purpose of improving the accuracy of the processing quality analysis of the baby stroller parts. Therefore, the present invention proposes a method for analyzing the processing quality of baby stroller parts based on visual inspection, so as to improve the analysis accuracy of the processing quality of baby stroller parts.

[0058] Embodiment 2: like Figure 2 , which is a functional module diagram of a system for analyzing the processing quality of baby carriage parts based on visual inspection provided by an embodiment of the present invention.

[0059] The processing quality analysis system 100 for baby carriage parts based on visual inspection according to the present invention can be installed in an electronic device. According to the functions implemented, the processing quality analysis system 100 for baby carriage parts based on visual inspection can include a quality key attribute analysis module 101, an image processing module 102, a feature optimization processing module 103, a component defect calculation module 104, a component accuracy calculation module 105 and a component quality evaluation module 106. The module according to the present invention can also be called a unit, which refers to a series of computer program segments that can be executed by an electronic device processor and can complete fixed functions, which are stored in the memory of the electronic device.

[0060] In this embodiment, the functions of each module / unit are as follows: The key quality attribute analysis module 101 is used to obtain the manufacturing process steps corresponding to the baby stroller components, query the manufacturing criteria corresponding to the manufacturing process steps, and analyze the key quality attributes corresponding to the baby stroller components based on the manufacturing criteria; The image processing module 102 is used to collect component preparation images corresponding to the baby stroller components in the preparation process, set component reference points corresponding to the baby stroller components based on the key quality attributes, and perform region extraction processing on the component preparation images in combination with the component reference points to obtain component observation images; The feature optimization processing module 103 is used to extract the component features to be measured corresponding to the baby carriage components from the component observation image, and perform optimization processing on the component features to be measured to obtain target features to be measured; The component defect calculation module 104 is used to retrieve the component reference image corresponding to the baby stroller component, and based on the component reference point, extract the reference feature to be measured from the component reference image, and calculate the component defect corresponding to the baby stroller component by combining the target feature to be measured and the reference feature to be measured, wherein the calculation of the component defect corresponding to the baby stroller component by combining the target feature to be measured and the reference feature to be measured includes: Performing vectorization processing on the target feature to be measured and the reference feature to be measured respectively to obtain a first feature vector and a second feature vector; Calculate the feature deviation between the target feature to be measured and the reference feature to be measured by combining the first feature vector and the second feature vector; The feature weight corresponding to the target feature to be measured is calculated, and the component defect degree corresponding to the baby carriage component is calculated by combining the feature weight and the feature deviation through the following formula: Among them, E represents the component defect degree corresponding to the baby stroller component, It represents the feature weight corresponding to the ath feature in the target feature to be tested. It represents the feature deviation corresponding to the ath feature in the target feature to be tested, a represents the feature sequence number of the target feature to be tested, r represents the number of features of the target feature to be tested, Indicates the maximum value of feature deviation; The component accuracy calculation module 105 is used to perform multiple tests on the dimensions corresponding to the components of the baby stroller to obtain component dimension data, determine the actual dimensions of the components corresponding to the components of the baby stroller based on the component dimension data, and calculate the component accuracy corresponding to the components of the baby stroller by combining the actual dimensions of the components and the component dimension data; The component quality evaluation module 106 is used to evaluate the component quality corresponding to the baby carriage component in combination with the component defect degree and the component accuracy, and generate a processing quality result corresponding to the baby carriage component based on the component quality.

[0061] In detail, each module described in the system 100 for analyzing the processing quality of baby carriage parts based on visual inspection described in the embodiment of the present application is used in the same manner as described above. Figure 1 The same technical means as the method for analyzing the processing quality of baby carriage parts based on visual inspection described in the specification and can produce the same technical effects will not be repeated here.

[0062] It is obvious to those skilled in the art that the present invention is not limited to the details of the above exemplary embodiments, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0063] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A method for analyzing the processing quality of baby carriage parts based on visual inspection, characterized in that: The method comprises: Obtaining a manufacturing process step corresponding to a baby stroller component, querying a manufacturing criterion corresponding to the manufacturing process step, and analyzing a key quality attribute corresponding to the baby stroller component based on the manufacturing criterion; Collecting component preparation images corresponding to the baby stroller components in the preparation process, setting component reference points corresponding to the baby stroller components based on the key quality attributes, and performing region extraction processing on the component preparation images in combination with the component reference points to obtain component observation images; Extracting the component features to be measured corresponding to the baby carriage component from the component observation image, and performing optimal processing on the component features to be measured to obtain target features to be measured; Retrieving a component reference image corresponding to the baby stroller component, and extracting a reference feature to be measured from the component reference image based on the component reference point, and calculating the component defect degree corresponding to the baby stroller component by combining the target feature to be measured and the reference feature to be measured, wherein calculating the component defect degree corresponding to the baby stroller component by combining the target feature to be measured and the reference feature to be measured includes: Performing vectorization processing on the target feature to be measured and the reference feature to be measured respectively to obtain a first feature vector and a second feature vector; Calculate the feature deviation between the target feature to be measured and the reference feature to be measured by combining the first feature vector and the second feature vector; The feature weight corresponding to the target feature to be measured is calculated, and the component defect degree corresponding to the baby carriage component is calculated by combining the feature weight and the feature deviation through the following formula: Among them, E represents the component defect degree corresponding to the baby stroller component, It represents the feature weight corresponding to the ath feature in the target feature to be tested. It represents the feature deviation corresponding to the ath feature in the target feature to be tested, a represents the feature sequence number of the target feature to be tested, r represents the number of features of the target feature to be tested, Indicates the maximum value of feature deviation; Perform multiple tests on the dimensions corresponding to the baby stroller components to obtain component dimension data, determine the actual dimensions of the components corresponding to the baby stroller components based on the component dimension data, and calculate the component accuracy corresponding to the baby stroller components by combining the actual dimensions of the components and the component dimension data; The component quality corresponding to the baby carriage component is evaluated in combination with the component defect degree and the component accuracy, and the processing quality result corresponding to the baby carriage component is generated based on the component quality.

2. The method for analyzing the processing quality of baby carriage parts based on visual inspection according to claim 1, characterized in that: The step of analyzing the critical quality attributes of the baby carriage components based on the manufacturing criteria includes: Performing text recognition on the preparation criteria to obtain a criteria text; Performing semantic analysis on the standard text to obtain the standard text semantics; Extracting key criterion text from the criterion text based on the semantics of the criterion text; Analyzing the criterion relevance between the key criterion texts, and extracting criterion labels corresponding to the key criterion texts; Performing attribute parsing on the key criterion text to obtain criterion attributes; The key quality attributes corresponding to the baby carriage components are extracted from the criterion attributes in combination with the criterion labels and the criterion associations.

3. The method for analyzing the processing quality of baby carriage parts based on visual inspection as claimed in claim 2, characterized in that: The analyzing the criterion relevance between the key criterion texts includes: Extracting key terms from the key criteria text, and standardizing the key terms to obtain standard key terms; Defining correlation factors in the standard key terms, and constructing a criterion correlation matrix corresponding to the key criterion text based on the correlation factors; Based on the criterion association matrix, calculating the criterion association degree between the key criterion texts; According to the criterion relevance, the criterion relevance between the key criterion texts is analyzed.

4. The method for analyzing the processing quality of baby carriage parts based on visual inspection according to claim 1, characterized in that: The step of performing region extraction processing on the component preparation image in combination with the component reference point to obtain a component observation image includes: In combination with the component reference point, performing region marking processing on the component preparation image to obtain a component marking preparation image; Performing region segmentation processing on the component marking preparation image to obtain a component region image; Performing noise reduction processing on the component region image to obtain a noise-reduced component image; The noise-reduced component image is digitally quantized to obtain a component observation image.

5. The method for analyzing the processing quality of baby carriage parts based on visual inspection as claimed in claim 4, characterized in that: The step of performing digital quantization processing on the noise reduction component image to obtain a component observation image includes: Constructing an image coordinate system corresponding to the noise reduction component image, and identifying a pixel coordinate value of each pixel in the noise reduction component image in the image coordinate system; Based on the pixel coordinate values, respectively calculating the pixel grayscale change rate and the change rate direction corresponding to each pixel in the noise reduction component image; Based on the pixel grayscale change rate and the change rate direction, edge extraction is performed on the denoised component image to obtain the component image edge; Perform edge fitting processing on the edge of the component image to obtain a component observation image.

6. The method for analyzing the processing quality of baby carriage parts based on visual inspection as claimed in claim 5, characterized in that: The step of calculating the pixel grayscale change rate and the direction of the change rate corresponding to each pixel in the noise reduction component image based on the pixel coordinate value includes: Based on the pixel coordinate values, calculating the horizontal grayscale change rate and the vertical grayscale change rate corresponding to each pixel in the noise reduction component image; Combining the horizontal grayscale change rate and the vertical grayscale change rate, the pixel grayscale change rate corresponding to each pixel in the noise reduction component image can be calculated by the following formula: Among them, B represents the grayscale change rate of each pixel in the noise reduction component image. Indicates the horizontal grayscale change rate, Indicates the vertical grayscale change rate; Combining the horizontal grayscale change rate and the vertical grayscale change rate, the change rate direction corresponding to each pixel in the noise reduction component image can be calculated by the following formula: Where D represents the direction of the rate of change corresponding to each pixel in the denoising component image, represents the inverse tangent function, Indicates the horizontal grayscale change rate, Indicates the vertical grayscale change rate.

7. The method for analyzing the processing quality of baby carriage parts based on visual inspection according to claim 1, characterized in that: The step of extracting the component features to be measured corresponding to the baby carriage component from the component observation image comprises: Performing contrast enhancement processing on the component observation image to enhance the observation image, and extracting observation color features corresponding to the enhanced observation image; Performing grayscale processing on the enhanced observation image to obtain a grayscale observation image, and extracting observation texture features corresponding to the grayscale observation image; Performing boundary detection on the enhanced observation image to obtain an observation boundary, and extracting boundary shape features corresponding to the observation boundary; The observed color features, the observed texture features and the boundary shape features are combined to generate component features to be measured corresponding to the baby carriage components.

8. The method for analyzing the processing quality of baby carriage parts based on visual inspection according to claim 1, characterized in that: The step of performing optimal processing on the component features to be measured to obtain target features to be measured includes: Performing threshold segmentation processing on the features to be tested of the component to obtain segmented features to be tested; Performing feature ablation processing on the segmented features to be measured to obtain ablated features to be measured; The characteristic Gini index corresponding to the ablation feature to be measured is calculated, and based on the characteristic Gini index, the ablation feature to be measured is optimized to obtain a target feature to be measured.

9. The method for analyzing the processing quality of baby carriage parts based on visual inspection according to claim 1, characterized in that: The calculating the component accuracy corresponding to the baby carriage component by combining the actual size of the component and the component size data comprises: The size mean corresponding to the component size data is calculated, and the size standard deviation of the baby stroller component in multiple tests is calculated by combining the size mean and the component size data using the following formula: Where H represents the standard deviation of the dimensions of the stroller parts in multiple tests. Indicates the size value corresponding to the i-th data in the component size data, i indicates the serial number corresponding to the component size data, and n indicates the number of data corresponding to the component size data. represents the mean value of size; Calculating the size tolerance corresponding to the stroller component based on the actual size of the component and the preset basic size of the component; Combining the size tolerance, the size mean, the actual size of the component and the size standard deviation, the component accuracy corresponding to the stroller component is calculated by the following formula: Among them, K represents the component accuracy corresponding to the baby stroller parts, represents the size tolerance of the bth component in the baby carriage. represents the mean size, Indicates the actual size of the bth component in the baby carriage. represents the size standard deviation of the bth component in the baby stroller, and b represents the serial number of the baby stroller component.

10. A processing quality analysis system for baby carriage parts based on visual inspection, characterized in that: The system comprises: A quality key attribute analysis module, used to obtain the manufacturing process steps corresponding to the baby stroller components, query the manufacturing criteria corresponding to the manufacturing process steps, and analyze the quality key attributes corresponding to the baby stroller components based on the manufacturing criteria; An image processing module, for collecting component preparation images corresponding to the baby stroller components in the preparation process, setting component reference points corresponding to the baby stroller components based on the key quality attributes, and performing region extraction processing on the component preparation images in combination with the component reference points to obtain component observation images; A feature optimization processing module is used to extract the component features to be measured corresponding to the baby carriage components from the component observation image, and perform optimization processing on the component features to be measured to obtain target features to be measured; A component defect calculation module is used to retrieve a component reference image corresponding to the baby stroller component, and based on the component reference point, extract a reference feature to be measured from the component reference image, and calculate the component defect corresponding to the baby stroller component by combining the target feature to be measured and the reference feature to be measured, wherein the calculation of the component defect corresponding to the baby stroller component by combining the target feature to be measured and the reference feature to be measured includes: Performing vectorization processing on the target feature to be measured and the reference feature to be measured respectively to obtain a first feature vector and a second feature vector; Calculate the feature deviation between the target feature to be measured and the reference feature to be measured by combining the first feature vector and the second feature vector; The feature weight corresponding to the target feature to be measured is calculated, and the component defect degree corresponding to the baby carriage component is calculated by combining the feature weight and the feature deviation through the following formula: Among them, E represents the component defect degree corresponding to the baby stroller component, It represents the feature weight corresponding to the ath feature in the target feature to be tested. It represents the feature deviation corresponding to the ath feature in the target feature to be tested, a represents the feature sequence number of the target feature to be tested, r represents the number of features of the target feature to be tested, Indicates the maximum value of feature deviation; A component accuracy calculation module, used for performing multiple tests on the dimensions corresponding to the components of the baby carriage to obtain component dimension data, determining the actual dimensions of the components corresponding to the components of the baby carriage based on the component dimension data, and calculating the component accuracy corresponding to the components of the baby carriage by combining the actual dimensions of the components and the component dimension data; The component quality evaluation module is used to evaluate the component quality corresponding to the baby carriage component in combination with the component defect degree and the component accuracy, and generate a processing quality result corresponding to the baby carriage component based on the component quality.

Images